Abstract Higher field (7T) MRI offers many potential advantages to clinical and scientific studies, including increased sensitivity and in many cases improved contrast. Demonstration of this potential in specialized studies has resulted in the proliferation of 7T and even 9.4T human scanners. There are, however, outstanding methodological challenges to bringing these systems from the research to the clinical arena. One of the most limiting issues is the ability to achieve uniform image contrast over the body. The source of this confound is the shortened RF wavelength in biological tissues leading to spatial inhomogenieties of the transmit B1 field, which, in turn lead to contrast variations in many sequences. In a conventional T1 weighted volume acquisition, the image contrast can change over an order of magnitude over the head at 7T. RF excitation in the presence of time-varying gradients offers the potential of multi-dimensional selective excitation. In this scheme, the excitation is shaped to mitigate the B1 pattern measured in the head. The principal challenge is to encode the 3D excitation in a short enough time to be useful for a wide variety of sequences, including 2D slice selective sequences as well as 3D (non-selective) acquisitions. We propose a development program to test the ability the spatially shaped pulses to mitigate B1 inhomogeniety in the head at 7T. Our first aim is to develop the RF transmit arrays needed to decrease the length of time spent encoding these pulses using the transmit SENSE method. Our second aim is to improve the methods used for calculating the accelerated pulses and test the limits of the transmit SENSE approach in the head at 7T, including high flip angle excitations. Our third aim is to build the methodology that will be needed to safely and accurately monitor local SAR levels for transmit arrays and assess the SAR penalty associated with constructive interference from the E fields of the multiple transmit channels